Google says your site is slow. Your visitors disagree. Who’s right?

Run your site through a speed tool and you may get a horror story: a performance score in the thirties, a headline metric of 40-plus seconds. Then check your analytics and find that real visitors load the page in a second and a half, buy things, and never complain. Both measurements are correct. They are answering different questions.

What the lab number is

Lighthouse — the engine inside most speed tools — loads your page once, on a simulated mid-range phone, over a throttled connection roughly equal to a weak 4G signal. It is deliberately pessimistic: it measures your site the way the least-equipped visitor on the worst network would experience it, with an empty cache, first visit, nothing warmed up.

What the field number is

Chrome also collects timings from real visits — actual people, actual networks, the last 28 days. That data (Google calls it CrUX) is what search ranking actually uses. If most of your visitors are on good phones in a city with fast coverage, your field numbers can be excellent while your lab number is dreadful.

Who is right, then?

Both — about different people. The field number describes the customers you already have. The lab number describes the customers you don’t: the ones on old phones and patchy networks who arrived once, waited, and left before your analytics ever saw them. Field data has a survivorship bias built in; people who give up don’t get measured much.

So read the pair like this. Good field, bad lab: your current audience is served, but you are turning away the marginal visitor — worth fixing if you want to grow beyond people with new phones. Bad field, any lab: stop reading and fix it; you are losing customers you can already name. Good both: close the tab, do something else.

Why we show you both

Our reports put Google’s own numbers first — the same Lighthouse and real-user data anyone can check — and then add what a single-page test cannot see: how every template on your site behaves, which of them carries the weight, and which pages share the problem. The one-page score is the baseline. The site-wide picture is where the fix actually lives.

The worst response to a scary score is a redesign. The right response is to find out which number is scary, for whom, and whether those people are the ones your business needs next.